📌 Feature & Improvement Tracker

Open NeptuneHub opened this issue 7 months ago • 0 comments

This issue is used to track feature requests and improvements to the project.

🛠️ Only maintainers will update this list.
📩 To request a feature, please open a new issue

[x] Natural language prompts for on demand playlist generation - #10
[x] Nearest Neighbor feature to show a song similar to another, or similar artist - #10 , #14
[x] Support ARM architecture - #20
[x] Add integration of new Media Server via API like Navidrome - #23
[x] Consider weekly mix playlists (max 4-5) instead of pre-generating many playlists - #14
[x] API to access to song information - #60
[x] Add reverse proxy support - #52
[x] Add library filtering to exclude unwanted albums from analysis and avoid playlists crossing library boundaries by default - #14
[ ] Display audio traits visually in the interface to give the user the possibility of input the playlist name based on them and/or give the possibility to add only some of the playlist generated - #10, #14
[ ] Include track duration as a feature in regression/clustering for better differentiation by genre - #14
[ ] Support multiple Media Server with one deployment - #132
[ ] Support multiple users

[x] Improve parameter documentation - #10
[x] Move advanced parameters to separate config section in the front-end - #10
[x] Simplify installation (especially for users unfamiliar with Docker/K3S) - #10
[x] Improve log to better show which parameter was used as input - #14
[x] UI improvements or directly create Jellyfin plugin as UI - #10
[x] Tune clustering algorithm (v0.3.1 already includes some improvements) - #10
[x] Try to minimize the case on which the song are not analyzed (example due to more than 2 channel, or codec version)
[x] similar Song: remove multiple version of the same song - #30
[x] similar song: avoid more than 3 song of the same artist - #30
[X] check integration with LMS by subsonic api
[x] use embedding vector to check if two or more song are duplicated - #30
[x] add icon to the Audiomuse-ai website - #33
[x] convert old .pb tensorflow model to new .keras and find a way to don't need to reload model at each inference
[ ] check the possibility for pagination on Similar song search
[ ] add automated test - #147
[ ] improve documentation - #176

[x] Remove playlist created with clustering with less then 10 songs
[ ] About the .env have a look to the TOP N moods configuration. User should be to change it and re-start the full analysis with the new parameter - #14

Don't wont list

✅ We’ll continue to update this issue as new requests come in and progress is made.

Jun 30 '25 12:06 NeptuneHub